The recent increase in utility research has provided improved methods for estimating the standard deviation of performance in dollars. Subjective estimates of an individual's overall worth to the organization allow the utility of various organizational interventions to be evaluated. However, this research does little to illuminate the dimensions underlying supervisory judgments of utility. The recent increase in utility research has provided improved methods for estimating the standard deviation of performance in dollars. Subjective estimates of an individual's overall worth to the organization allow the utility of various organizational interventions to be evaluated. However, this research does little to illuminate the dimensions underlying supervisory judgments of utility. The policies underlying judgments of overall worth were captured to a substantial degree, with cross-validated R² values ranging from .46 to .69. A unit weighting scheme was applied to the six predictors, resulting in r² values that exceeded the cross-validated R² derived from regression analyses. This substantial predictability of utility judgments provided the capacity to generalize utility information from a sample of jobs to the population of interest. Analyses comparing validity-based and utility-based clustering schemes explored the degree of convergence between the two approaches to classifying jobs. These analyses indicated that there was some overlap, with validity information being useful in establishing broad categories of jobs associated with similar utility-relevant attributes. At the same time, these analyses demonstrated that the two approaches were not equivalent. Implications of this research are discussed, and several possible directions for future research are noted. It is suggested that such policy capturing procedures can enhance our understanding of judgments of overall worth, and expand the knowledge base upon which organizational decisions are made. / Ph. D.